More specifically, the present invention relates to a method in which a main drive train of a vehicle to be tested is connected to at least one load machine. If the main drive train propels multiple axle drives, individual load machines connected by means of side shafts may brake and propel independently from one another in order to simulate driving resistances and/or acceleration moments or delay moments. Control of the load machine(s) is usually based on a simulation calculation of the vehicle, down to the individual tires if desired. In particular, the measurement of the rotational movement refers to a measurement of torque and/or speed taken in this conjunction. Examples for methods and/or corresponding dynamometers which the invention intends to improve have already been described in AT 508 031 B1 or Bauer R., “New Methodology for Dynamic Drive Train Testing” in Proceedings of the Symposium on International Automotive Technology, pp. 1-6, 2011, reference to which is made in particular with regard to control of the load machines in the context of methods and/or a chassis dynamometer of the present type.
In recent years, the technology of these dynamometers for the drive train has developed to such an extent that on a so-called full-vehicle chassis dynamometer the test object is subject to the exact same loads as on the road. This allows “vehicle application on the dynamometer” by which, instead of time- and cost-intensive driving tests on a test track, the same tests may be performed on the dynamometer (see Pillas J., Kirschbaum F., Jakobi R., Gebhardt A., Uphaus F.: “Model-based load change reaction optimization using vehicle drivetrain test beds”, Tagungsband des 14. Internationalen Stuttgarter Symposiums, pp. 857-867, 2014).
Here, vehicle applicators criticise that the accelerations of the vehicle as a response to actions of the test object (accelerating, braking, changing gears etc.) on the dynamometer are naturally different from actual driving on roads. The subjective feel (“seat-of-the-pants feel”) of the vehicle applicators, in particular, is thus different on the dynamometer as during actual driving on roads.
Naturally, the vehicle on the dynamometer cannot perform the same movements as on the road. In addition, the vehicle must have a good mechanical connection to the load machines (and/or roller machines). If one intended to simulate the acceleration via tilting, as in a flight simulator, this would require moving the entire trip rail including the vehicle and the load machines, which is made impossible by the large mass.
In a different context, namely dynamometers for examining the impact of forces and torques acting onto an undercarriage via wheels and wheel suspensions, EP 2 602 602 A1 already suggests providing translational actuators and connecting the drive assemblies to the wheels of the test object via double-joint shafts. In this case, a test load, for example in the form of a longitudinal force, is applied to the wheels by the actuators. In this type of dynamometer, however, the translational actuators are exclusively reconstructing predetermined values for distance or force. At the most, a coordination not specified in more detail of the predetermined values for the test load and a drive torque and/or drive speed is provided.
In other known dynamometers of this type having translational actuators, e. g. according to DE 102 12 255 A1, DE 10 2012 101 613 A1 or EP 0 094 570 B1, a rotation of the wheels and/or side shafts of the drive train is virtually irrelevant.
An acceleration of the vehicle as a response to actions of the test object on the dynamometer comparable to the vehicle's behaviour during actual driving on roads is not provided in the known methods and dynamometers.
As a consequence, it is the object of the invention to provide a method of the initially mentioned type which allows a subjective feel which is close to reality, i. e. similar to a driving test on a test track, for a person situated within a vehicle. Preferably, the behaviour of the vehicle is simulated close to reality during accelerating, braking, changing gears etc. of the respective test object.
To achieve this object, it is provided, according to the invention, that in the method of the initially mentioned type a corresponding longitudinal acceleration is determined from the measured rotational movement and the actuator is controlled in accordance with the longitudinal acceleration which is determined based on the measurement. According to this and according to the invention, it is provided that in a chassis dynamometer of the initially mentioned type a control device, which is connected to the measuring device and to the actuator, is configured to determine a corresponding longitudinal acceleration from a measured rotational movement of the wheel obtained from the measuring device and to transmit a control signal to the actuator in accordance with the determined longitudinal acceleration. Control of the actuator, which is formed by a linear motor, for example, is not performed on the basis of an arbitrary predetermined test load or a merely calculated simulated acceleration, but dynamically, based on a measurement of the rotational movement of a wheel or a drive train of the vehicle, i. e. in particular based on a direct or indirect measurement of the torque carried out by the wheel, and preferably in real-time (“on-line”). In this way, a realistic response of the test object to the vehicle control performed by the vehicle (accelerating, braking, changing gears etc.) may be simulated in an advantageous manner. In general, a “vehicle application on the dynamometer” is useful and possible only if the test object responds to the physically caused impact on the wheel and/or the drive train in the right way—i. e. according to the real test object—since it is only here that changes, for example of the motor control device, have an impact on the acceleration being calculated on-line from the vehicle model and being transferred into the test object by the actuator, so these changes have an impact on the subjective feel of the vehicle applicator. On the other hand, a subjective feel is basically possible if predetermined values for distance or force are used, as is the case in the prior art, but in general this feel does not correspond to the actual and/or real behavior of the vehicle so vehicle application is hardly useful.
The exact longitudinal acceleration the vehicle would perform on the road can only be used to a very limited extent on the dynamometer since, in a full-throttle scenario, for example, this would result in a distance of over 100 m covered in only a few seconds. As has been revealed for longer tests, for controlling the actuator the determined longitudinal acceleration may be converted into a subjective acceleration in order to shorten a covered translational distance with respect to a distance resulting from the determined longitudinal acceleration. The conversion into a subjective acceleration exploits the fact that in general the acceleration perceived by a person deviates from the physical acceleration. Because of this, it is preferable to choose a conversion for approximating a subjectively correct impression of acceleration while minimising the distance covered by the test object.
For example, it has been proven favorable for low-frequency portions of the determined longitudinal acceleration to be reduced during conversion of the acceleration. This approach is based on the knowledge that only or mainly high-frequency portions of the acceleration are relevant for subjective human perception. This means that relative abrupt changes of acceleration are perceived primarily while consistent acceleration or slow changes are perceived to a significantly lower extent—if at all. In this context, portions in the range of 1-2 Hz and below, in particular, are considered low-frequency portions.
Furthermore, it is advantageous if the determined longitudinal acceleration is modified by a high-pass filter, for example a high-pass filter designed according to Bessel or Butterworth or preferably a first-order high-pass filter, for the conversion into the subjective acceleration, with a time constant of the high-pass filter preferably being between 0.01 and 1 second, in particular about 0.1 seconds. Such a filtering of the longitudinal acceleration may be implemented relatively efficiently and allows responding in real-time and avoiding noticeable latencies compromising the subjective perception. According to this, it is advantageous for the control device in the present chassis dynamometer to comprise a high-pass filter for the determined longitudinal acceleration, for example a high-pass filter designed according to Bessel or Butterworth or preferably a first-order high-pass filter, with a time constant of the high-pass filter preferably being between 0.01 and 1 second, in particular about 0.1 seconds.
In order to further reduce the distance actually covered by the vehicle within the dynamometer, it is favourable for the displacement of the actuator to be set to a constant nominal position, in particular in the middle of an intended travel of the actuator. In this way, multiple accelerations, which are subsequent but temporally interrupted and/or consistent over certain segments, may be simulated in the same direction and/or with the same algebraic sign within an overall shorter translational path, in particular within the same translational segment. In this case, the translational segment is covered according to the acceleration, multiple times and in rapid succession in the same direction, wherein between the accelerations the vehicle is moved back to the starting position, the constant nominal position if applicable, relatively slowly and thus unnoticeably by the position controller. Basically, the nominal position may be selected such that the accelerations to be expected may be put into effect. In the middle of the intended travel, accelerations in both directions (forwards or backwards and/or accelerating or braking) are possible likewise. If different accelerations and/or accelerations of different velocities are expected depending on the direction, the nominal position may be adapted according to these expectations, of course.
A particularly simple and reliable control may be obtained if the displacement of the actuator is controlled by a position controller, preferably having a rise time between 0.05 and 5 seconds, in particular having a rise time of about 0.5 seconds. By the same token, it may be provided that in the proposed chassis dynamometer the control device comprises a position controller for controlling the displacement of the actuator, preferably comprising a rise time between 0.05 and 5 seconds, in particular about 0.5 seconds. A common PID controller or a cascade controller, for example one having a velocity controller formed by a PI controller and an upstream displacement controller formed by a P controller, may be used as the position controller.
In cooperation with the position control circuit it is favorable for a portion proportional to the subjective acceleration to act on the position control circuit as a disturbance, in particular if a portion proportional to the subjective acceleration is added to a control variable of the position control circuit. Preferably a force, which corresponds to the subjective acceleration, with a known vehicle mass and a portion associated to a single actuator in case of multiple actuators, is added to a force predetermined for the actuator, for example, by a position controller. This means that an acceleration portion of the position control circuit overlies the subjective acceleration performed by the actuator, i. e. the low-frequency portions of the determined longitudinal acceleration are replaced by the control variables of a position controller. Preferably, the position controller is not set as tight as possible so the additional force corresponding to a subjective acceleration may also be implemented by the actuator.